Navigating the Boundaries of Emergency Department Care: Addressing the Medical and Social Needs of Patients Who Are Homeless

Abstract

Objectives. We sought to understand interpersonal and systems-level factors relevant to delivering health care to emergency department (ED) patients who are homeless.

Methods. We conducted semistructured interviews with emergency medicine residents from 2 residency programs, 1 in New York City and 1 in a medium-sized northeastern city, from February to September 2012. A team of researchers reviewed transcripts independently and coded text segments using a grounded theory approach. They reconciled differences in code interpretations and generated themes inductively. Data collection and analysis occurred iteratively, and interviews continued until theoretical saturation was achieved.

Results. From 23 interviews, 3 key themes emerged: (1) use of pattern recognition in identifying and treating patients who are homeless, (2) variations from standard ED care for patients who are homeless, and (3) tensions in navigating the boundaries of ED social care.

Conclusions. Our study revealed practical and philosophical tensions in providing social care to patients in the ED who are homeless. Screening for homelessness in the ED and admission practices for patients who are homeless are important areas for future research and intervention with implications for health care costs and patient outcomes.

More than 1.5 million people experience homelessness in the United States every year.¹ Homelessness is associated with high levels of chronic illness and morbidity,^2–4 unmet health needs,^5–8 and difficulties in accessing primary care.^9,10 People who are homeless use the emergency department (ED) at higher than average rates,¹¹ even when compared with other low-income populations.^12–14 Furthermore, people who are homeless are disproportionately represented among frequent users of the ED.^12,15,16 For example, an analysis at 1 urban hospital found that 38% of patients with 5 or more ED visits per year were homeless.¹⁵

Most studies pertaining to homelessness and ED use have focused on predictors of ED use, identifying such correlates as medical comorbidities, mental illness, substance abuse, crime, and food insecurity.^14,17–22 Fewer studies have examined basic descriptive information about ED visits by patients who are homeless. One national survey showed that ED patients who were homeless had triage urgencies and hospital admission rates comparable to those of other patients but were more likely to have arrived by ambulance.^23,24 A paucity of research, however, has examined in any detail the ED care received by patients who are homeless. Significant knowledge gaps exist about potential challenges related to ED care of patients who are homeless, including the response to their medical and social needs and any differences in their care compared with that for other patients.

Improving ED care for patients who are homeless without first understanding the unique interpersonal and systems-level challenges in providing such care will be difficult. To this end, we aimed to explore providers’ perceptions of ED care for patients who are homeless using semistructured interviews with emergency medicine resident physicians (“residents”) as key informants. We sought to gain meaningful insight into the multifaceted processes of delivering health care to patients who are homeless in the ED to guide improvements in current practice and generate hypotheses that can be further explored in future research.

METHODS

We conducted a qualitative study using semistructured, 1-on-1 interviews with emergency medicine residents.²⁵ We chose a qualitative approach given the expected complexity of residents’ experiences with and feelings toward patients who are homeless, involving residents’ own emotions as well as an intricate interplay of interpersonal and systems-level issues in the ED, which would be difficult to capture quantitatively, particularly for a potentially sensitive topic involving a stigmatized population.^26–28

Setting and Participants

We selected emergency medicine residents as key informants because they are generally the primary physician caregivers in teaching hospital EDs, and they interact directly with multiple members of the ED care team, including nurses, social workers, attending physicians, and support staff. Throughout their training, residents acquire a rich and unique insight into the full scope of ED processes of care.

We recruited residents from 2 large (50–60 residents) northeastern emergency medicine residency programs. One program is located in New York City, and residents work at an urban safety-net hospital serving many homeless patients (Site A). The other program is located in a medium-sized northeastern city, and residents work at an academic medical center serving a variety of patients ranging from those who are privately insured to a sizable homeless population (Site B). We selected these programs to capture a wider breadth of resident experiences and inform a more complete understanding than would be possible from interviewing residents at 1 site alone.

We sampled residents with diverse representation from program site and postgraduate training year (PGY) to ensure that a range of experiences was included, though the study was not designed to formally compare differences in residents by training year. Complete lists of all residents at both sites were obtained. To avoid selection bias, we used a random-number generator to select a random sample of residents stratified by training year and site. We initially selected 24 residents with a plan to recruit and collect data until we achieved theoretical saturation (the point at which new concepts no longer emerge from additional interviews).^29,30 Selected residents were invited to participate via e-mail, with the chance to win a $100 gift card offered as an incentive to participate.

Data Collection

We obtained verbal informed consent from all participants. Interviews were conducted in a private location by a single investigator from February through September 2012. We used a standard interview guide (see the box on the next page) including 7 core questions with probes to ensure that key concepts were covered while allowing for flexibility in the interview structure and for new concepts to emerge.²⁸ We selected initial interview guide concepts on the basis of a review of the prior literature and our own experience working with patients who are homeless in the ED. Interviews were digitally recorded and professionally transcribed, and transcriptions were rechecked against the audio recordings for accuracy. Personal or patient-identifying information was removed from transcripts before analysis.

Interview Guide

1.	Could you just tell me a little bit about your experience caring for homeless patients here at [institution name]?
	Broad question to begin to allow residents to identify the most salient issues.
2.	Tell me about the last patient or a recent patient you saw who was homeless?
	Residents asked to provide specific examples in this and other questions to allow themes to be identified by concrete experiences.
3.	3a) Is caring for homeless patients different in any way from caring for other patients in the ED?
	3b) Are there particular things that you like or don’t like about taking care of homeless patients in the ED?
4.	4a) How does your knowledge that a patient is homeless affect your clinical decision making?
	4b) How have you learned about how to care for homeless patients?
5.	How do you feel when the next patient to be seen is homeless?
6.	Lastly, are there any experiences that stand out in your mind that are related to taking care of a homeless patient during your residency so far?
7.	What else that I have not asked about yet do you think would be important to talk about?

Data Analysis

A core team of 3 researchers (K. M. D., S. P., F. E. V.) with diverse content-relevant expertise reviewed and coded all transcripts independently. They met regularly, after each set of 3 to 4 interviews had been individually coded, to discuss and reconcile differences in code interpretations and work toward a final code structure. This core coding team consisted of an emergency physician who recently completed residency training and has significant previous work and service experience with homelessness outside the ED, a senior emergency medicine faculty member who has worked in the ED for 20 years but has not worked with homeless populations outside the ED, and a public health student with prior experience working with homeless individuals accessing a syringe-exchange program. All core coding team members had formal training in and experience with qualitative research. Other research team members (A. A. V., L. A. C., M. G.) participated in coding and provided input on the code structure at key points in the analysis.

Transcripts were reviewed line by line in the grounded theory tradition. Grounded theory, as defined by Glaser and Strauss,^30–33 allows codes to emerge organically from the interview text rather than from a predefined list. This method maximizes the chance that codes and themes will be grounded in study participants’ experiences rather than reflecting researchers’ preconceived notions.^31,34 The constant comparison method was used to identify new codes and refine existing ones while adjusting the code structure accordingly.^28,34 We conducted data collection and analysis iteratively in waves of 3 to 4 interviews to allow for refinements in the interview guide as needed and for the evaluation of theoretical saturation, which was determined by group consensus.

Once interviews were completed and a final code structure was developed, 2 investigators (K. M. D., S. P.) independently recoded 3 transcripts using the final code structure. The 2 investigators had more than 90% agreement in coding. Therefore, the remaining transcripts were recoded by a single investigator (K. M. D.) using the final code structure. We loaded final transcripts into ATLAS.ti (ATLAS.ti Scientific Software Development GmbH, Berlin, Germany) for thematic analysis and organization.

RESULTS

Of the 24 residents selected, 23 responded and agreed to participate. One resident did not respond to invitation e-mails and was replaced by another randomly selected resident from the same site and training year. Theoretical saturation was reached after 23 interviews, so interviews were stopped at that point. The mean interview length was 33 minutes (range = 17–54 minutes). Participant characteristics appear in Table 1. Most residents reported having a small amount of prior experience with patients who were homeless during medical school rotations but little other significant prior experience with homeless populations.

TABLE 1—

Characteristics of Emergency Medicine Residents Interviewed: February–September 2012

Characteristics	Site A (n = 11)	Site B (n = 12)
Gender
Male	6	6
Female	5	6
Postgraduate training year
1	2	3
2	3	3
3	3	3
4	3	3

Three recurrent and unifying themes emerged: (1) the use of pattern recognition in identifying and treating patients who are homeless, (2) variations from standard ED care for patients who are homeless, and (3) tensions in navigating the boundaries of ED social care (see the box on p. S358).

Main Themes and Supporting Concepts: February–September 2012

Use of pattern recognition in identifying and treating patients who are homeless

• Difficulty in identifying which patients are homeless

• Overlap with substance abuse and mental illness

• Clinical hypervigilance vs clinical complacency

Variations from standard ED care for patients who are homeless

• Hospital admission decisions

• Unique external burdens and barriers (e.g., time pressures, difficulty obtaining medications)

• Personal factors (e.g., communication, hygiene)

Tensions in navigating the boundaries of ED social care

• Social needs addressed by ED for individual patients (e.g., food, shelter) and as social safety net (e.g., patients brought to ED for public intoxication)

• Resident boundaries: how much is “my job”?

• Organizational boundaries: tension between ED’s role and limitations in addressing social issues

Note. ED = emergency department.

Using Pattern Recognition to Identify and Treat Homeless Patients

Pattern recognition, a heuristic strategy commonly used by emergency physicians to act quickly with limited information, involves acting on stereotypes or biases about groups of patients, based in part on generalizing from previous patients cared for in the ED. One resident explained, “I profile my patients. I assume everybody does on some level. If you don’t, you’re either lying or probably really inefficient” (PGY-1, Site B).

Residents described challenges in identifying ED patients who were homeless. Most reported that they did not routinely ask all patients if they were homeless. Instead, they often relied on pattern recognition, suspecting that patients were homeless when they fit certain stereotypes. One resident noted, “A lot of times it’s hygiene; you’ll notice people whose nails are dirty or people are wearing multi-layers of clothing” (PGY-3, Site A). Several residents were surprised to learn that a given patient who did not have the typical appearance associated with homelessness was indeed homeless. Participants found defining the homeless patient population to be complicated by the substantial overlap between homelessness and substance abuse and mental illness. Although most recognized intellectually that different types of homeless patients existed, the most prominent profile of homelessness in the ED was the patient who was alcoholic and chronically homeless: “When you say ‘a homeless person,’ I picture the drunk, homeless person” (PGY-3, Site A).

Residents also described using pattern recognition to determine patient acuity or illness severity in a busy ED. Emergency physicians commonly use such pattern recognition to quickly determine whether patients are sick or not sick, but making such a distinction appeared more challenging with patients who were homeless, with residents wrestling with being hypervigilant versus complacent. On one hand, residents reflected that seeing so many patients who were homeless and simply intoxicated or looking for food or shelter could lull them into assuming that all patients who were homeless were not acutely ill:

Sometimes you don’t have time to really delve into what’s going on with them … and it seems like you remember the last time they were here and all they wanted was food and they somehow survived to come back for more food, so you just kind of have these biases especially for people that you see all the time. (PGY-1, Site A)

On the other hand, participants also feared “the one homeless guy who ended up having something real [a critical illness]” (PGY-1, Site A) and noted that heightened vigilance was required for patients who were homeless. One resident explained, patients who were homeless fell into the “don’t get yourself burned” category (PGY-1, Site B). Another elaborated,

Because they’re homeless and often intoxicated people want to blow them off … I try and make sure I think about each one … because there’s often something to find that could be really significant that if I just blew off as a drunk homeless person I would have missed. (PGY-3, Site B)

Variations From Standard Care for Homeless Patients

The most consistently reported influence of homelessness on ED care involved decisions about whether to admit patients to the hospital or discharge them from the ED. All participants reported being more likely to admit homeless patients to the hospital than patients who had similar illnesses but were not homeless:

You are more likely to be a bit more conservative … like if it’s a small cellulitis you would send someone home with antibiotics, I don’t think I would send a person like that to a shelter knowing that no one is going to see it. Things I might teeter on the verge of admitting or not admitting, I would be more likely to admit someone who’s homeless. (PGY-4, Site B)

Several other external barriers to standard ED care existed for patients who were homeless. For example, residents reported rushing their clinical care to help patients arrive at shelters by a certain hour for intake. Other practical considerations included patients’ susceptibilities to different illnesses; unsafe, unclean, or crowded living environments; and difficulty obtaining medications or accessing follow-up care. As one resident said, “It’s just a different ballgame. There’s so many barriers to care that it’s very difficult and you have to put in a lot more work” (PGY-4, Site A).

Certain patient characteristics additionally affected standard ED care. Communication with some patients who were homeless could be challenging, particularly with those who were intoxicated or had active psychosis. A few residents also noted that whereas most ED patients are helped to undress fully to facilitate physical examination, patients who were homeless were less likely to be undressed because of their poor hygiene. These challenges were exacerbated by pressure to quickly treat patients in crowded EDs, which came into conflict with the extra time often required to ideally evaluate and develop a care plan for patients who were homeless.

Tensions in Navigating the Boundaries of Social Care

Participants discussed addressing not only medical needs of patients who were homeless but also social needs, including shelter and food. One resident explained,

If they wanna hang out in bed 7 for a couple hours and eat a sandwich and get some sleep and be warm … for me, it’s an easy patient that I can take care of, it’s somebody I can make feel better. (PGY-1, Site A)

Particularly in cold weather or late at night, residents admitted to allowing patients who were homeless to sleep in the ED. They also referred patients to shelters, either with the help of a social worker or by themselves. Participants noted a variety of other nonmedical services that EDs provided, including clothing, transportation, and showers. These social services were typically provided alongside more traditional medical services. In some cases, however, social services were the only ones needed during an ED visit:

A guy was living in his car in the winter and came in. He says that his car ran out of gas and he was getting cold. I think “ran out of gas” was the chief complaint. [We] gave him blankets and let him stay in the department ’til morning. (PGY-4, Site B)

Participants reported that, although patients could present to the ED for social needs on their own accord, they were often brought in involuntarily. One resident noted that when concerned citizens reported a person sleeping in a public place, the local response was for that person to be transported to the ED. Others noted the large number of patients brought to the ED by ambulance for public intoxication:

I’d say the biggest population of homeless that we see are alcoholics. They’re there because [our state] requires people who are publically intoxicated to come in to an emergency department—they don’t have drunk tanks in this state. (PGY-2, Site B)

Participants described tensions in determining the boundaries of how far ED care should extend into the social realm. Individual residents embraced social care to varying degrees as part of their job. Some felt that addressing social needs was an integral component of ED care: “That’s a large part of what we do, but I also feel like it’s very rewarding” (PGY-4, Site A). Others hewed more closely to their medical roles: “I can’t be a social worker also” (PGY-3, Site B). Residents noted that time spent dealing with social needs detracted from time that could be spent caring for acutely ill patients.

In addition to navigating this ill-defined boundary individually, residents’ comments reflected a tension extending to the philosophy and roles of EDs in general. Nearly all pointed to limitations in how much the ED could do to help patients who were homeless. They felt that although EDs could help with medical and even discrete social needs of individual patients, a larger systems-level fix was needed that fell outside the purview of EDs:

I think the real difference lies—if we wanna break the cycle—somewhere outside of the emergency room, somewhere in the government or social, public health. (PGY-4, Site B)

Participants felt particularly ineffectual with patients who were both alcoholic and chronically homeless: “When you’re working a shift in the emergency department,” said one, “You’re not gonna fix their chronic homelessness” (PGY-4, Site A).

DISCUSSION

Using emergency medicine residents as key informants, we identified 3 key themes related to ED care of patients who are homeless: (1) the use of pattern recognition in identifying and treating patients who are homeless, (2) variations from standard ED care for patients who are homeless, and (3) tensions in navigating the boundaries of ED social care.

Consistent with prior empirical research, residents identified a significant overlap of homelessness with substance abuse and mental illness,^16,23 speaking to the need for enhanced ED connections with substance abuse and mental health treatment programs. For example, ED alcohol screening and intervention programs^35,36 could be expanded to consider special needs of homeless populations and could be paired with more intensive programs for patients who are chronically homeless and have severe alcohol use problems.^37,38 At the same time, because ED patients who are homeless may not have the typical appearance that providers associate with homelessness (i.e., intoxicated, unkempt), providers may underidentify homelessness in certain subgroups, such as women or those who are newly homeless. Indeed, a few residents commented that, especially during the economic recession and housing foreclosure crisis, they had been surprised to learn that certain patients (e.g., a former banker) were homeless. These findings suggest the need for more uniform homelessness screening and documentation procedures in the ED. Multiple ways exist in which knowledge of a patient’s homelessness status can or should affect ED care, including considerations around diagnosis, disposition, and appropriate discharge plans. Furthermore, many people access the ED as a first-stop site soon after becoming homeless.³⁹ Therefore, with accurate screening the ED could become an important site for early interventions such as rapid rehousing programs.⁴⁰

Our second theme, variations from standard ED care for patients who are homeless, has important implications with the finding that residents have a lower threshold for admitting patients who are homeless to the hospital. This finding should be confirmed with future research. Communities and hospitals could consider collaborating on safe and potentially cost-saving alternatives to hospitalizations such as medical respite programs that provide specialized services for patients who are too sick to be on the streets or in a traditional shelter but not sick enough to need an inpatient hospital bed.^41,42

Finally, we discovered tensions in navigating the boundaries of ED social care. Previous research has found high levels of social needs among ED patients⁴³ and has provided insights into some of the ways in which EDs meet those social needs.^44–48 In addition to meeting the social needs of individual patients on a case-by-case basis, residents in our study also noted the ED’s role as a de facto 24-hour shelter and sobering center. Thus, the ED serves not only as a medical safety net but also as a social safety net.⁴⁹ EDs are unique among health care settings because their doors remain open 24 hours a day, 7 days a week, and they have a federal mandate⁵⁰ to serve everyone requesting treatment. Unlike other social service agencies, EDs cannot turn patients away on the basis of eligibility criteria or capacity considerations, and thus they fill a crucial yet underfunded and underrecognized role in providing social care. Many of these same observations were presented in an article published more than 10 years ago, in which Gordon⁴⁹ discussed challenges and opportunities for EDs to serve as social welfare institutions. Yet in the intervening decade, relatively little research has examined the ED’s provision of social care. Important questions remain about what should be the appropriate boundaries of social care offered in EDs, how social care fits into the ED’s mission, and whether providing social care for patients who need it can improve patient health and reduce future ED utilization. Multicenter studies could start by examining variations in social care delivery among different EDs and best practices for addressing patients’ social needs.

Limitations and Strengths

Our study focused on residents from 2 urban residency programs. Residents’ experience in rural or suburban EDs may differ from that of residents in this study. We interviewed residents as key informants because they regularly and closely interact with patients who are homeless and with other care providers in the ED. Additional insight could be gained by speaking directly to other members of the ED care team or to patients themselves. Finally, social desirability bias may have influenced participants’ responses. We minimized such bias by using well-accepted techniques such as establishing rapport, ensuring confidentiality, and encouraging participants to give examples.²⁸

Despite its limitations, our study represents a significant contribution to understanding the care that patients who are homeless receive in the ED, providing new insight into the limited qualitative research that has been conducted with ED nurses and patients.^48,51,52 Our study has uncovered questions for future exploration regarding the difficulties in identifying patients who are homeless in the ED, how homelessness affects ED care, and tensions surrounding social care provision in the ED. Future quantitative research is warranted to confirm some of our study findings.

Conclusions

The ED is an important and common site of health care for patients who are homeless. Using qualitative methods, our study revealed unique insights into ED care of patients who are homeless that have implications for both future research and immediate practice changes. By examining ED care for patients who are homeless, we also highlighted the broader issue of the ED’s role in providing social care. More research is needed to better understand such social care and how to best provide it.